The direct cost of stroke that currently stands at around US$35 billion dollars annually represents only the tip of the iceberg. Unemployment, missed work days, and premature mortality extend this figure, as costs attributed to these indirect expenditures equate to a further US$68.5 billion dollars. The cost analysis that connoted this conclusion was presented by Tarun Girotra (University of New Mexico, Albuquerque, USA) at the International Stroke Conference (ISC; 5–8 February, Honolulu, USA).
Reporting that nearly two thirds of the total cost associated with stroke, at US$103.50 billion per year—when analysed using non-institutionalised US civilian populations—were found to be indirect costs, Girotra highlighted that the strokes occurring in the 35–64 age range were to blame for the bulk of this economic loss. Furthermore, patients at 45-64 years of age incurred more annual direct costs than those aged 65–79 years, despite having a lower prevalence of stroke.
“The cost of illness analysis primary consists of two costs”, said Girotra, addressing the ISC audience, “direct costs consist of expenditures relate to hospitalisation, nursing home care, healthcare provider visits, medications and medical durables. But equally important are the indirect costs; the loss of productivity from unemployment, underemployment and premature mortality”.
Pointing to previous limitations of prior stroke cost of illness analyses, Girotra put forward that many solely focused on direct costs and were not inclusive of all age groups, while others used older databases without national reach. In light of this gap in the literature, Girotra and colleagues decided to perform a comprehensive cost of illness analysis for stroke in the US, incorporating all age groups.
Data for the current analysis was sourced from the Medical Expenditure Panel Survey (MEPS) from 2003 to 2014. “For those of you who are unfamiliar with the database, it is a multistage sample representative of the noninstitutionalised US civilian population,” Girotra explained. He also noted that the subjects’ insurance provider and medical provider was available, which was used to corroborate the information obtained from the subjects.
For the direct costs—defined as the sum of the indirect out-of-pocket payments and payments issued by insurance—Girotra and colleagues looked at expenditures from the database pertaining to inpatient hospital stay, outpatient care, prescription medications and more. Indirect cost components were obtained through the US National Vitals Statistics database, where employment status, the number of missed work days and premature mortality was calculated.
According to Girotra, a chi-square test to compare baseline characteristics between stroke and non-stroke patients was initially implemented. Next, a two-part econometric model was used to ascertain the adjusted incremental expenditure for patients with stroke versus non-stroke patients. “For this, we implemented a probit model to test the probability of observing a zero versus positive expenditures”, noted Girotra. A generalised linear model with gamma distribution and log link was carried out, while Girotra said that all incremental expenditures were controlled for gender, ethnicity, education, insurance status, census region, income, marital status and CCI.
Addressing the indirect costs, Girotra said that a negative binomial regression model was used in the adjusted model to estimate the difference between missed work days. While a logistic regression model was used to estimate the probability of full year employment, an economic model first proposed by Grosse et al—known as the Present Value of “one life”—calculated indirect costs resulting from premature mortality. According to Girotra, this model includes market and non-market (household) lifetime productivity for different age groups in both genders, and utilises the American Time Use Survey (ATUS). Finally, Girotra also highlighted the fact that the US dollar values were inflated to the 2016-dollar value using the Consumer Price Index obtained from the US Bureau of Labor Statistics.
Findings
Of the 253,235,052 civilian noninstitutionalised US population that were identified, just over 8 million (8,101,159; 32%) had stroke. Weighted samples of 10,175 stroke patients and 314,694 non-stroke patients were compared.
Girotra acknowledged that as far as baseline variables were concerned, the overall stroke prevalence was higher in patients 65 years or older, females, white and non-hispanic black, non-married, high school graduate, publically insured participants, Midwest and South regions, as well as poor and low-income categories.
“We were able to get a total incremental direct cost [of stroke] per single patient; of US$4,317”, remarked Girotra. Of this average figure, which ranged from US$3,828 to US$4,807, Girotra noted that just under 50%, US$1,920, were related to inpatient costs.
However, through looking at the overall annual incremental direct cost of stroke, Girotra reported that the total was US$35 billion. He acknowledged this himself and his colleagues observed an interesting finding when they stratified cost by age: “The age group of 45–64 years—although had a lower stroke prevalence and a lower stroke population, incurred a greater direct cost of US$14.5 billion, compared to the 65–79 group, which incurred US$12.3 billion”.
After indirect costs were adjusted for variables, Girotra said that the annual wage difference between stroke survivors and non-stroke patients was -US$14,254 (95% CI: -US$16,023, -US$12, 485); a difference of which they attributed to unemployment or underemployment. “When we looked at the annual wage difference and expanded it to the number of stroke patients we had in the country, the annual loss was US$38.1 billion”, commented Girotra.
In addition, stroke survivors had 4.2 more missed work days (per year) compared to the non-stroke patients, while their risk of unemployment was 60% higher compared to non-stroke patients. Subsequent study into premature mortality, through utilising Grosse et al’s Present Value of one life econometric model, yielded a total loss of US$30.4 billion. “The interesting finding here is that the age group of 35–65 incurred an economic loss of US$14.6 billion—roughly half of the total US$30.4 billion dollars”, Girotra exclaimed.
Through aggregating the total annual costs spent on stroke patients compared to non-stroke patients in the USA, Girotra reported that the total incremental direct cost (US$35 billion), plus the total indirect cost from underemployment (US$38.1 billion) and premature mortality (US$30.4 billion) equates to $103.5 billion.
“There are certain limitations of these kinds of studies,” noted Girotra, “the MEPS database includes noninstitutionalised citizens only, meaning the population living primarily in nursing homes are not surveyed. This is a large chunk of stroke survivors which have a lot of economic impact on society in general”. Furthermore, according to Girotra, the authors were unable to calculate indirect costs associated with loss of productivity from unemployment and absenteeism of family members acting as caregivers. Paediatric strokes are also not routinely picked up as adult strokes are, noted Girotra, which may also underestimate the total annual cost of stroke.
Despite these shortcomings, Girotra maintained that through this nationally representative database, the direct cost of stroke represented only the tip of the iceberg, with two thirds of the total cost attributed to indirect costs. Addressing the fact that stroke still remains a highly preventable disease, Girotra said: “There is a significant financial incentive to increase community outreach programmes to control stroke risk factors and improve stroke care delivery models to increase acute stroke treatment”.
